Results in Engineering 9 (2021) 100196

Contents lists available at ScienceDirect

Results in Engineering
journal homepage: www.editorialmanager.com/rineng/Default.aspx

CFD simulation of a hyperloop capsule inside a closed environment

Federico Lluesma-Rodríguez a, b, Temoatzin Gonz

alez b, Sergio Hoyas a, *
a
Instituto Universitario de Matematica Pura y Aplicada, Universitat Politecnica de Valencia, 46022, Valencia, Spain
b
Zeleros Global S.L. (Hyperloop), 46024, Valencia, Spain

A R T I C L E I N F O A B S T R A C T

Keywords: One of the most restrictive conditions in ground transportation is traveling through a tunnel at high speed. In
CFD those conditions waves are propagated, increasing the pressure upstream the object, and so, the drag compared to
Evacuated tube train (ETT) the open flow case. Although this drawback is mitigated with larger tunnels, another proposed solution is to
Train
decrease the pressure inside the tunnel. In this paper it is demonstrated that the drag coefficient is almost
Vacuum tube
invariant with the pressure conditions. This effect allows, not only to have smaller tunnels with respect to the
existing for standard trains, but also to enhance the speed of the train without increasing its aerodynamic losses.

1. Introduction
Electric propulsion is the future in transportation [13]. This is the
reason why research in trains to overcome current speeds is performed.
In that sense, aerodynamics is not a minor issue if higher velocity want to
be achieved. Apart from other limits, such us the catenary contact [14] or
the wheel-rail contact [2], the aerodynamics is present in the ground
transportation systems as an strong limitation, specially when focusing in
tunnels. Although this problem has been widely studied [4,5,7,11], this
paper is focused on what occurs in a closed and controlled environment,
or Evacuated-Tube Train (ETT).
The aim of this study is to reproduce the conditions of a train inside a
closed tunnel for different blockage ratios, static pressure and speeds. To
reduce the cost of the process avoiding expensive tests [12], CFD simu-
lations have been performed. The main output is the aerodynamic drag
under those conditions. The size analysed for the tunnel start from the
typical one for a High-Speed Rail (blockage ratio of 0.23 [1,10]) up to
0.75.
2. Methodology
The geometry of the train is based on a model of the Transrapid
Maglev train used in Ref. [3], whose maximum speed is 505 km/h [6].
The model, shown in Fig. 1, has a length of L ¼ 51:7 meters and a height
of h ¼ 3:7 meters.
The capsule is embedded into a cylindrical fluid domain. A radial
plane can be seen in Fig. 1. In order to avoid wave reflection, the inlet and
outlet of the tunnel are placed 10 times away the length L of the train.
The blockage ratio β is defined as the quotient between the cross
sectional area and the tunnel. Once this parameter is fixed, the height of
the tunnel, ht , is derived. The different boundaries on the domain are:
 Inlet: mass flow inlet based on train speed and reference pressure. The
inlet static temperature is set to 288.15 K. Total pressure inlet is
checked in results.
 Tube: moving and adiabatic wall, with the same speed as the train in a
ground reference frame.
 Outlet: pressure outlet, equal to reference pressure.
 Vehicle: static and adiabatic wall.
 Axis: rotation axis to convert the 2D domain into an axilsymmetric
one.
The spatial scheme used is a second order one, while the turbulence
model is k  ω SST as in Ref. [3]. The solver employed is ANSYS Fluent.
The mesh is a hybrid one, formed using the blocks shown in Fig. 2. All
of them are structured but two, marked with an u. These blocks are used
to adapt the boundary layer mesh to the near field domain avoiding bad
quality elements.
Three meshes have been proposed to perform the mesh sensibility
study (see Table 1), using the drag coefficient cD . The set up chosen for
this study is the most restrictive one, having the highest speed (700 km/
h), pressure (1 atm) and β (0.75). This is the case where waves are
stronger, due to the larger energy on the inlet flow.
The final mesh chosen is the medium size one. The reason is because
the difference in the drag with respect to the fine mesh is less than 1%,
which is also lower than the difference with respect to the coarse one.

* Corresponding author.
E-mail address: serhocal@mot.upv.es (S. Hoyas).

https://doi.org/10.1016/j.rineng.2020.100196
Received 13 October 2020; Received in revised form 1 December 2020; Accepted 7 December 2020
2590-1230/© 2020 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
F. Lluesma-Rodríguez et al. Results in Engineering 9 (2021) 100196

Table 3
Drag for different sources or boundary conditions (β ¼ 0:5).
Vref pref Drag ([3]) Drag (1) Drag (2)

700 km/h 0.022 atm 83 081 N 148 426 N 36 768 N

500 km/h 0.021 atm 44 694 N 063 446 N 13 561 N
Fig. 1. Numerical domain for the case.

than the reference value, while case 2 is the opposite. The main reason is
the larger upstream pressure when imposing the mass flow, as discussed
below in Table 4.
This means that the actual value of drag is highly dependent on the
upstream boundary condition. In the real case, due to wave propagation
in a closed environment where more than one train is circulating, this
upstream pressure varies continuously. The consequence is that the drag
Fig. 2. Blocking chosen for the mesh.
is not only a matter of the actual speed of the train, but also of the po-
sition of the train in the tunnel and what has happened previously, which
cannot be reproduced with the presented simplification of the problem.
Table 1
Meshes used in the sensibility study. As a representative average result, the boundary condition chosen is
mass flow. This condition is the most restrictive, and it represents the
Mesh Elements yþ cD Error [%]
case in which all the mass flow goes through the channel between the
Coarse 96045 4.63 16.539 – tunnel and the train, which is more realistic in a tunnel with closed walls.
Medium 171947 0.83 16.473 0.40
This is a relevant difference with respect to an open tunnel, the one
Fine 366079 0.41 16.477 0.02
commonly studied by other authors [8,9], where leaks of mass flow can
happen.
Due to wave propagation the flow is unsteady. However, to reduce the In any case, an effect that always occurs with the train traveling at this
computational cost of the simulation, a comparison between a steady and speed inside the tunnel is that it increases its pressure upstream (pinlet )
an unsteady case (with time step 0.012 s) has been performed. and decreases the speed of the flow (Vinlet ), as seen in Table 4.
An unsteady Riemann solver was also tested. A priori, this solver is Note that, almost independently of the train speed Vref , the actual
the one recommended for highly compressible cases, where compression flow speed upstream is similar in all cases Vinlet . This is a consequence of
waves appear. This result, along with the ones from the steady compar- the blockage of the channel between the train and the tunnel. Regarding
ison, is collected in Table 2. The set up for the case is the same as for the the pressure, imposing mass flow increases considerably the upstream
mesh sensibility one. pressure, a 50% at 500 km/h and a 100% at 700 km/h. When pressure is
Pressure based Steady solver is used for the rest of the cases, ac- imposed, this effect also appears, but is less effective.
cording to the results shown in Table 2. Note that the difference between In any case, this increase in the inlet pressure leads to the drag dif-
this solver and the most used one (Riemann) is less than 1%, while is ferences seen in Table 3.
noticeable faster. To further detail what occurs with the pressure, the pressure coeffi-
cient along the tunnel is represented in Fig. 3. As previously seen, there is
3. Results and conclusions a considerable increase in pressure on the front face of the train, larger as
the tunnel gets smaller (larger β). However, the difference with the
The drag coefficient is based on the wall speed, the tunnel reference reference pressure is not remarkable.
pressure, and the cross sectional area of the train. It is important to note that downstream the train there is a strong
One of the uncertainties of this case is the boundary condition on the depression, followed by an oscillatory behavior due to the obliques shock
inlet and outlet. For the outlet, the pressure is fixed to the same value as waves still present on the tunnel. Finally, there is a sudden compression
the reference pressure. However, for the inlet, two different conditions of the flow to accommodate the outlet pressure. Note that for the largest
have been tested: tunnel (β ¼ 0:20) there are no waves downstream, as the flow is not
chocked when passing through the small gap between the train and the
 Impose mass flow (case 1), computed using the reference values and tunnel.
the area of the tunnel: These flow patterns are better seen when representing the contours of
the Mach number in Fig. 4. There, the presence of shock waves due to the
m_ ¼ ρref π h2t Vref : (1) high speed flow on the gap is evident.
A parametric study using different values for the velocity (500 and
700 km/h), pressure (0.01, 0.1 and 1 atm) and blockage ratio (from 0.2 to
 Impose total pressure (case 2), computed using the reference pressure
0.75) has been performed. The results are shown in Fig. 5. There is low
and speed:
dependency of the drag coefficient with the pressure, which means that
!γ1
g
the drag scales linearly with that parameter. In the figure, it is also shown
2
γ  1 Vref the big dependency with the blockage ratio. If tunnel has β > 0:2, the
pt ¼ pref 1þ : (2)
γ γRTref drag starts to grow exponentially. Remember that the common value for
Results are shown in Table 3. Note that case 1 provides higher drag
Table 4
Table 2 Velocity (in km/h) and pressure (in atm) for β ¼ 0:5.
Different solvers used.
Imposed Vref pref Vinlet pinlet
Solver Drag Error [%]
Pres 700 0.022 373 0.026
Riemann 16422718 – Mass 700 0.022 374 0.043
Pressure based Transient 16408200 0.088 Pres 500 0.021 325 0.022
Pressure based Steady 16407199 0.095 Mass 500 0.021 358 0.030

2
F. Lluesma-Rodríguez et al.
Fig. 3. Coefficient of pressure along the tube for different cases (700 km/h).
Fig. 4. Contours of Mach number for β ¼ 0:75, 700 km/h, and 1 atm.
Fig. 5. Drag coefficient for the parametric study.
a High Speed Rail is 0.23, making it evident why the size of the tunnels is
not smaller.
Finally, the speed affects values of the cD curve. This is why the drag
coefficient is far from being considered constant with the speed on this
range.
Results in Engineering 9 (2021) 100196
Focusing on an example, if taken the drag of the train at 700 km/h in a
common tunnel (β ¼ 0:2), the cD is 0.64 at atmospheric pressure. This
blockage ratio means that the tunnel has a radius 124% higher than the
one from the train. The same drag can be achieved when reducing the
pressure to 0.1 atm and using β ¼ 0:5 (cD ¼ 6:50), which means that the
tunnel would only required to be 41% larger in terms of the radius with
respect of the train.
If future ground transport want to overcome 500 km/h, a new concept
where ambient pressure can be decreased must be proposed. This, allows
to decrease the drag of the capsuled linearly with this parameter, leading
to acceptable air resistances even at very high speeds and small tunnels.
Declaration of competing interest
The authors declare that they have no known competing financial
interests or personal relationships that could have appeared to influence
the work reported in this paper.
Acknowledgment
SH and FLR work have been supported by project RTI2018- 102256-
B-I00 of Mineco/FEDER. FLR is partially funded buy MCIU under grant
DI-17-09616.
References
[1] Adif, Tunel de Guadarrama. Technical report. Adif, URL: http://www.adifaltaveloci
dad.es/es_ES/infraestructuras/doc/tunelguadarrama.pdf, 2007.
[2] M. Givoni, F. Ltd, Sgm, Development and impact of the modern high-speed train: a
review, Transport Rev. 26 (2006) 593–611.
[3] T.K. Kim, K.H. Kim, H.B. Kwon, Aerodynamic characteristics of a tube train, J. Wind
Eng. Ind. Aerod. 99 (2011) 1187–1196, https://doi.org/10.1016/
j.jweia.2011.09.001.
[4] H. Lee, C. Park, J. Lee, Improvement of thrust force properties of linear synchronous
motor for an ultra-high-speed tube train, IEEE Trans. Magn. 47 (2011) 4629–4634,
https://doi.org/10.1109/TMAG.2011.2158584.
[5] T. Li, X. Zhang, Y. Jiang, W. Zhang, Aerodynamic design of a subsonic evacuated
tube train system, Fluid Dynam. Mater. Process. 15 (2019) 121–130, https://
doi.org/10.32604/fdmp.2020.07976.
[6] Maglev Board, Transrapid Maglev shanghai, URL: https://www.maglevboard.net/e
n/facts/systems-overview/transrapid-maglev/transrapid-maglev-shanghai.
[7] M. Mossi, S. Sibilla, Swissmetro : Aerodynamic Drag and Wave Effects in Tunnels
under Partial Vacuum, 2002.
[8] A. Orellano, Aerodynamics of High Speed Trains, 2008.
[9] R.S. Raghunathan, H.D. Kim, T. Setoguchi, Aerodynamics of high-speed railway
train, Prog. Aero. Sci. 38 (2002) 469–514, https://doi.org/10.1016/S0376-
0421(02)00029-5.
[10] RENFE, AVE serie 102/112, URL: https://www.renfe.com/viajeros/nuestros_trenes
/aves102_ficha.html, 2020.
[11] X. Zhang, Y. Jiang, T. Li, Effect of streamlined nose length on the aerodynamic
performance of a 800 km/h evacuated tube train, Fluid Dynam. Mater. Process. 15
(2019) 67–76, https://doi.org/10.32604/fdmp.2020.07776.
[12] M.S. Escarti-Guillem, S. Hoyas, L.M. García-Raffi, Rocket plume URANS simulation
using OpenFOAM, Results in Engineering 4 (2019), https://doi.org/10.1016/
j.rineng.2019.100056.
[13] P.K. Senecal, F. Leach, Diversity in transportation: why a mix of propulsion
technologies is the way forward for the future fleet, Results in Engineering 4
(2019), https://doi.org/10.1016/j.rineng.2019.100060.
[14] F. Alkam, I. Pereira, T. Lahmer, Qualitatively-improved identified parameters of
prestressed concrete catenary poles using sensitivity-based Bayesian approach,
Results in Engineering 6 (2020), https://doi.org/10.1016/j.rineng.2020.100104.